Reference Enterprise Architecture

[Chris04]

Can anyone direct me to a Proxmox reference enterprise architecture document?

I'm looking for a clear document that explains how Proxmox should be deployed for a scalable Enterprise solution - discussing ideal design, recommended node configuration, networking, storage, etc.

I am looking to migrate from our HyperV environment where we host 700 Windows + 200 Linux VMs using 1,800TB of storage, and are looking for the proper way to approach this. All new hardware will be acquired for this environment, so everything is on the table.

I've spent days searching - and have not been able to find anything thorough that I can take to management.

Dell posted this cursory Proxmox architecture document, but it's light on details:
https://infohub.delltechnologies.co...t-on-dell-powerflex/network-architecture-195/

Here is one for VMware - basically, I'm looking for the Proxmox version of this:
https://i.dell.com/sites/content/bu...VMware-vSphere-Reference-Architecture-SMB.pdf

The Proxmox Admin Guide is good for the background information, and how to implement with some 'rules of thumb' for deploying Ceph and the like, but it stops short of what the ideal design should look like.

The Proxmox Documentation is also good, but again, it does not discuss the details of what the ideal design should be.

Various forum posts talk about this, but sadly most ended in either 'take our paid training class', or 'it depends'. Dell, HP, IBM, VMware and even Microsoft publish reference documents that clearly state how their solutions should be deployed - having a clear reference for Proxmox would be very helpful.

Thank you,

Chris

 

[FrankList80]

Thats the point. The VCF and their validated design is the result of major Presales work. Our beloved Proxmox is not there yet (the core technological yes, all the stuff 20.000 VMWARE FTE did, not). This is an investment you have to cover (e.g. hiring externally Consultancy). P.S: you could try an LLM for a rough sketch .... but then you enhance with consultancy work ...

---

[Virtualization Support for SME and NGOs | DLI Solutions ]

 

[UdoB]

Maybe you could ask them via https://shop.proxmox.com/contact.php ?

That's what I would do with such a large system to plan...

 

[bl1mp]

but sadly most ended in either 'take our paid training class', or 'it depends'

Well from my perspective, that is the burden and advantage of Open Source.
There is no necessity to limit the list of possible solutions and divide them in "right" or "wrong" implementations. The question is what is the problem to solve and is it an useful solution, or is there a more suitable approach. To judge about that, it is necessary to understand the actual task.
(e.g. how many disaster domains, required tiering, existing storage (types) and backup solutions, budgets etc.)
and than find a solution.
This includes effort, which needs a lot of experience to build that knowledge.
It also takes time for people to match this knowledge with the specific customer requirements.
Otherwise you will always have a bad customer experience.
Abstracting away complexity to focus on advantages is part of the process of building a product.

To give an example, from my perspective, of two ideal solutions:
PVE runs as well on my laptops as well as on our clusters, but the configuration is quite different.
By the way in my opinion, the documents linked are not even as rich as the pve wiki on building usable clusters.

So from my perspective it would be helpful to follow the advice of taking paid classes and consultation.
This also has the bonus feature, that it is an invest in an Open Source product, which directly strengthens the availability for your business, avoiding inconvenience you might have with your current virtualization vendor.

Another approach could be to get clear your requirements (disaster domains, tiering, storages etc.)
and either ask for support on specific questions or start researching aiming in that direction.

All the best,
Lucas

 

[Johannes S]

If you are planning on rebuying the hardware a Ceph+PVE HCI Cluster+ two (one local for fast backup / restore and one offsite e.g. on a vserver or rented dedicated server in a datacenter ) Proxmox Backup Servers might fit the bill, the basic requirements are in the wiki:
https://pve.proxmox.com/wiki/Deploy...r#_recommendations_for_a_healthy_ceph_cluster

Depending on your workload and budget dedicated storage hardware with support for ProxmoxVE would be one option too (e.G. Blockbridge but @bbgeek17 knows more than me about his companys products ).
In any case I would highly recommend you to do an official training and get help from an experienced Proxmox Partner (so Gold or Silver) otherwise the chance is high that you mess something up right from the start. At least this is something several of the pros here said nearly everytime when a question like yours appeared: "If you cheap out on the storage or network hardware or do an early misstake in your architecture, it's quite likely it will bite you back later when it's much more difficult to fix it". Since I I myself use ProxmoxVE only in my homelab I can't say anything from my own experience but I have no reason to distrust them.
And although I love ProxmoxVE if you already have the WindowsServer licences and also mainly run WindowsVMs it might be actually more affordable (in terms of costs, migration burden etc) to stay in Hyper-V (at least for the WindowsVMs), why exactly are you considering a migration? I remember an offline discussion where some independent virtualization consultant said, that for a Windowsonly or mostly-Windows shop he would still recommend to stay with Hyper-V (since it's usually already payed, admins are already familiar how it works and because you won't run into driver issues or stuff like that due to Hypervisor and Guest OS are both coming from MS).